Data Science Final Presentation

12/09/2024

Ian Palmer

Project 1 - Part 1: Lisa’s Vegetable Garden

  • Dataset: TidyTuesday’s May 28, 2024 release, “Lisa’s Vegetable Garden”.

    • Records from Lisa Lendway’s 2020 and 2021 vegetable gardens.
    • Includes planting, harvesting, and spending data.

  • Plan: Explore harvest trends in 2020 by calculating average monthly harvest weights and visualizing seasonal patterns.

  • Skills Developed:

    • Developed skills in data wrangling, date manipulation, and visualization with the tidyverse.

Visulization

# A tibble: 5 × 2
  month      avg_weight
  <date>          <dbl>
1 2020-06-01       78.8
2 2020-07-01      219. 
3 2020-08-01      473. 
4 2020-09-01     1128. 
5 2020-10-01      959.

Project 2 - Part 2: Carbon Emissions

  • Dataset: TidyTuesday’s May 21, 2024 release, “Carbon Emissions”.

    • Historical emissions data from 1854–present.

  • Plan: Summarize average annual carbon emissions and visualize trends over time.

  • Devloped Skills:

    • Strengthened skills in data summarization, trend analysis, and visualization using the tidyverse.

Visulization

summary_data <- emissions |>
  select(year, total_emissions_MtCO2e) |>
  group_by(year) |>
  summarize(ave_emissions = mean(total_emissions_MtCO2e, na.rm = TRUE))

Project 2: Netflix Title Analysis

  • Dataset: TidyTuesday’s April 20, 2021 release, “Netflix Titles”.
    • Metadata on Netflix movies and TV shows, including release years, titles, and associated details.
  • Plan: Explore trends in Netflix’s content catalog:
    • Analyze the distribution of movies vs. TV shows.
    • Examine release year trends and common title keywords.
  • Devloped Skills:
    • Learned to apply string manipulation and visualization techniques to text-heavy data sets with tidyverse.

Visualization 1

netflix_type_count <- netflix |>
  count(type)

Visualization 2

netflix_year_count <- netflix |>
  count(release_year) |>
  arrange(desc(n))

Visualization 3

Project 3: NCAA Home-Court Advantage

  • Dataset: Kaggle March Madness Data (2008–2024, excluding 2020).
  • Plan: Test the null hypothesis of no difference in win percentages between home and away games using permutation testing.
  • Developed:
    • Learned to simulate random outcomes, perform hypothesis testing, and visualize results.

Permutation and Null Hypothesis Test

#Calculate the Observed Difference in Median Win Percentage
observed_diff <- combined_median_data |>
  group_by(location) |>
  summarise(median_win_pct = median(median_win_pct, na.rm = TRUE)) |>
  summarise(diff = diff(median_win_pct)) |>
  pull(diff)

cat("Observed Difference in Median Win Percentage (Home - Away):", observed_diff, "\n")
Observed Difference in Median Win Percentage (Home - Away): 31.16884 
# Define Permutation Function (Shuffling location within each team)
calculate_permutation <- function(data) {
  data |>
    group_by(TEAM) |>
    mutate(location = sample(location, replace = FALSE)) |>
    group_by(location) |>
    summarise(median_win_pct = median(median_win_pct, na.rm = TRUE)) |>
    summarise(diff = diff(median_win_pct)) |>
    pull(diff)
}
#Permutation Test
num_permutations <- 10000
perm_results <- map_dbl(1:num_permutations, ~ calculate_permutation(combined_median_data))

# Calculate the two-sided p-value
p_value <- mean(abs(perm_results) >= abs(observed_diff))
cat("Two-Sided P-value:", p_value, "\n")
Two-Sided P-value: 0

Visulization

Project 4: SQL Analysis of WAI Auditory Data

  • Goals: Recreate Figure 1 from Voss (2019) and analyze frequency absorbance trends by age.
  • Approach: Use SQL to query the WAI Database and R for visualization.
  • Findings:
    • Strengthened understanding of combining SQL and R for efficient analysis and clear communication of results.

SQL Query 1: Recreating Voss Graph


SELECT 
  Measurements.Identifier,
  PI_Info.AuthorsShortList,
  Measurements.Instrument,
  Measurements.Frequency,
  AVG(Measurements.Absorbance) AS MeanAbsorbance,
  CONCAT(PI_Info.AuthorsShortList, ' et al. N=', 
         COUNT(DISTINCT CONCAT(Measurements.SubjectNumber, Measurements.Ear)), ', ', Measurements.Instrument) AS Legend_Label
FROM Measurements
JOIN PI_Info ON Measurements.Identifier = PI_Info.Identifier
WHERE Measurements.Identifier IN ('Abur_2014', 'Feeney_207', 'Groon_2015', 'Lewis_2015', 'Liu_2008', 'Rosowski_2012', 'Shahnaz_2006', 'Shaver_2013', 'Sun_2016', 'Voss_1994', 'Voss_2010', 'Werner_2010')
  AND Measurements.Frequency >= 200  -- Apply frequency filter in SQL
GROUP BY Measurements.Identifier, Measurements.Instrument, PI_Info.AuthorsShortList, Measurements.Frequency;

Visulization 1: Recreating Voss Graph

SQL Query 2: Deeper look into Hunter 2016 Study

SELECT 
  Subjects.AgeCategoryFirstMeasurement AS AgeCategory, 
  Measurements.Frequency, 
  AVG(Measurements.Absorbance) AS MeanAbsorbance
FROM Measurements
JOIN Subjects ON Measurements.SubjectNumber = Subjects.SubjectNumber
WHERE Measurements.Identifier = 'Hunter_2016' 
  AND Measurements.Frequency >= 200  -- Apply frequency filter in SQL
GROUP BY Subjects.AgeCategoryFirstMeasurement, Measurements.Frequency;

Visulization 2: Deeper look into Hunter 2016 Study

Thanks For Listening!